CVE-2026-74682

In the Linux kernel, the following vulnerability has been resolved:

ALSA: usb-audio: fix OOB write on Type II inbound URBs

data_ep_set_params() sizes each URB transfer buffer before it adds the Format Type II transfer delimiter:

u->packets = urb_packs; u->buffer_size = maxsize * u->packets;

if (fmt->fmt_type == UAC_FORMAT_TYPE_II) u->packets++; /* for transfer delimiter */ u->urb = usb_alloc_urb(u->packets, GFP_KERNEL);

buffer_size is computed from the pre-increment packet count and never recomputed, so for a Type II endpoint the buffer is one packet short of the packet count the URB is built with.

prepare_inbound_urb() then lays out one iso frame per packet and never consults buffer_size:

offs = 0; for (i = 0; i < urb_ctx->packets; i++) { urb->iso_frame_desc[i].offset = offs; urb->iso_frame_desc[i].length = ep->curpacksize; offs += ep->curpacksize; }

urb->transfer_buffer_length = offs; urb->number_of_packets = urb_ctx->packets;

The last descriptor therefore points one packet past the end of the transfer buffer, where the host controller writes device data on every inbound transfer. prepare_silent_urb() and prepare_playback_urb() bound their fill loops by ctx->buffer_size, so only capture is affected.

fmt_type comes from the device's audio streaming descriptors, so any device advertising a Type II capture format hits this once userspace sets hw_params on the stream.

KASAN on 7.2.0-rc5 (arm64) with a dummy_hcd/raw-gadget device, one report per inbound transfer:

BUG: KASAN: slab-out-of-bounds in dummy_timer Write of size 64 at addr ffff0000186171c0 by task cons02/166 __asan_memcpy dummy_timer hrtimer_run_softirq Allocated by task 166: usb_alloc_coherent snd_usb_endpoint_set_params The buggy address is located 0 bytes to the right of allocated 64-byte region [ffff000018617180, ffff0000186171c0)

Compute buffer_size after the delimiter packet has been accounted for, and bound the fill loop by buffer_size, as prepare_silent_urb() already does on the outbound side. This grows every Type II URB allocation by one maxsize packet.

Discovered by XBOW, triaged by Baul Lee <baul.lee@xbow.com>